Intel loses its Lustre – Chipzilla bins own-brand HPC file system

Intel's decided to stop offering its own version of the Lustre file system – the code beloved of high-performance computing types because it's handy for managing exabyte-scale storage spanning colossal Linux clusters.

Trish Damkroger, an Intel veep and general manager for technical computing initiatives – that's HPC in Intel-Speak, this week sent an email to partners and customers in which she says:

Starting today, Intel will contribute all Lustre features and enhancements to the open source community. This will mean that we will no longer provide Intel-branded releases of Lustre, and instead align our efforts and support around the community release.

The email, sighted by The Register, goes on to say: “These changes are designed to increase Intel’s involvement in the community and to accelerate technical innovation in Lustre. For the community as a whole this will mean easier access to the latest stable Lustre releases and an acceleration of the technical roadmap.

“Intel will be heavily involved in producing the Lustre community releases and expects there to be the same high level of quality that customers have come to expect from the Intel-branded releases,” Damkroger adds. “Furthermore, Intel will be publishing community maintenance releases for a designated Long Term Stable (LTS) release so that users can easily access an up to date 'latest and greatest'.”

That LTS release will be Lustre 2.10, due in May 2017. Users of Intel-branded will be offered help to upgrade to 2.10. Intel is also continuing its Lustre support business.

Chipzilla's already set the ball rolling by popping its Hadoop Adapter for Lustre and HPC Adapter for Mapreduce/Yarn(HAM) onto GitHub. Damkroger promises more code will appear there soon.

Does this matter?

Lustre users probably won't much care. The community edition proceeds apace and will continue to benefit from Intel's input. Indeed, Chipzilla has good reasons to keep Lustre humming so its HPC customers continue to love its Xeons and keep their roaming eyes off those cute little Power chips and ARMs. And let's not forget that the HPC community is not exactly short of big brains knowledgeable about deep-level file system wonkery, so should be able to consume and integrate whatever Intel keeps tossing into the Lustre code base. Long story short, the file system does not look to be on death row, even if it has lost a prominent commercial distribution.

Symbolically, however, this seems a substantial decision. Intel's already bailed from an OpenStack collaboration and killed its Developer Forum event. Binning its own Lustre distribution and those other changes starts to look a bit like getting rid of some dead wood and/or spring cleaning. That the company is also throwing the kitchen sink at artificial intelligence further suggests Intel is engaging in a significant mid-course correction. ®